News Intel Partners With Baidu on Neural Network Processor for Training

[Admin]

Intel has partnered with Baidu on the development of its upcoming 16nm Nervana Neural Network Processor for Training (NNP-T), both on software and hardware.

Intel Partners With Baidu on Neural Network Processor for Training : Read more

 

[JayNor]

"Cascade Lake’s DL Boost, Intel’s new instructions to double or triple the performance of FP16 or INT8 AVX-512 vector code."

DLBoost doesn't do anything for fp16. The Cooper Lake chip is supposed to be the one that adds the bfloat16 avx512 vector operations.

The DLBoost/VNNI AVX512 additions are all fused multiply-add for int8 or int16. The DLBoost intrinsics are all described here, pretty clearly.

https://software.intel.com/sites/la...3,2214,2215,2222,2223&avx512techs=AVX512_VNNI